Journal of Biomechanics 35 (2002) 585–594

Postural control during lifting

J. Kollmitzer*, L. Oddsson1, G.R. Ebenbichler, J.E. Giphart1, C.J. DeLuca1

Department of Physical Medicine and Rehabilitation, University of Vienna, W .ahringer G .urtel 18-20, 1090 Vienna, Austria

Accepted 28 November 2001

Abstract

Any voluntary motion of the body causes an internal perturbation of balance. Load transfer during manual material handling

may increase these perturbations. This study investigates effects of stance condition on postural control during lifting. Nineteen

healthy subjects repeatedly lifted and lowered a load between a desk and a shelf. The base of support was varied between parallel

and step stance. Ground reaction force and segmental kinematics were measured. Load transfer during lifting perturbed balance. In

parallel stance postural response consisted of axial movements in the sagittal plane. Such strategy was accompanied by increased

posterior shear forces after lift-off. Lifting in step stance provided extended support in anterior/posterior direction. The postural

control mechanisms in the sagittal plane are less complex as compared to parallel stance. However, lifting in step stance was

asymmetrical and thus accompanied by distinct lateral transfer of the body. Lateral shear forces were larger as compared to parallel

stance. Both lifting techniques exhibit positive and negative aspects. We cannot recommend either one as being better in terms of

postural control. r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Perturbation; Balance; Manual material handling; Center of mass (COM); Center of pressure (COP)

1. Introduction

Ergonomic lifting advice for safe material handling ofloads commonly includes the use of a wide stance andperformance of the lift with bent knees and a straightback to decrease loads on the spine. Such guidelines arebased on knowledge from biomechanical studies thathave investigated mechanical aspects of spinal loading(Chaffin, 1987; McGill et al., 1996; Fathallah et al.,1998; Chaffin et al., 1999) and focused to reduce the riskof low back injury within the lumbar disks. Surprisingly,only few studies have investigated postural controlstrategies during lifting or performing a materialhandling task (Toussaint et al., 1997; Toussaint et al.,1998; Oddsson et al., 1999). Specific lifting guidelines arelacking, that take into consideration the postural controlmechanisms necessary for safe material handling.

Due to the multi-link structure of the varioussegments of the human body, any voluntary movement

will impose a perturbation of equilibrium. Theseperturbations increase when the movement is performedwith an added load such as lifting an object. Forexample, front-loading of the body will cause a shift ofthe system center of mass (COM) forward andanticipatory postural adjustments will be applied thatcounteract the upcoming perturbation (Brown andFrank, 1987). These involuntary ‘‘automatic’’ move-ments are smoothly incorporated into our movementrepertoire to ensure accurate and harmonious motion(Massion and Gahery, 1979; Oddsson, 1990; Timmannet al., 1994; Ioffe et al., 1996). These postural synergiesare triggered prior to the onset of voluntary movementsand appear to be flexible and task specific (Dietz et al.,2000; Bouisset and Zattara, 1987; Diener et al., 1983).Thereby, a certain voluntary movement may beassociated with different automatic postural adjust-ments depending on the context of the task (Crennaet al., 1987; Pedotti et al., 1989; Shiratori and Latash,2000; Cordo and Nashner, 1982; Hodges et al., 2000).When voluntary movements are performed underunstable situations these postural adjustments becomemore complex, and involve more muscles, higheractivation levels (Oddsson, 1989, 1990; Aruin et al.,1998), and/or different activation strategies such as

*Corresponding author. Tel.: +43-1-40400-4294; fax: +43-1-40400-

5281.

E-mail address: josef.kollmitzer@akh-wien.ac.at (J. Kollmitzer).1Experiments performed at: NeuroMuscular Research Center,

Boston University, 19 Deerfield Street, 4th Floor, Boston, MA

02215, USA.

0021-9290/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.

PII: S 0 0 2 1 - 9 2 9 0 ( 0 1 ) 0 0 2 3 8 - X

co-contraction (Humphrey and Reed, 1983; Oddsson,1990; Nielsen and Kagamihara, 1992; Accornero et al.,1997; Cresswell et al., 1994).

This study investigates effects of stance condition onpostural adjustments during lifting. We hypothesizedthat a larger base of support in step stance would reducecomplexity of postural control. In addition wedeveloped a model with postural adjustmentsabsent. Comparison between the passive reactionforces of the model and data recorded from subjectsmight provide better insight into the postural controlstrategies.

2. Methods

2.1. Subjects

Nineteen healthy subjects (5 female/14 male)participated in the study. Their age ranged from 20 to45 years (mean 27a/S.D. 6a); body weight from 51 to93 kg (mean 71 kg/S.D. 11 kg); and body height from164 to 191 cm (mean 179 cm/S.D. 8 cm). Before theexperiment, all subjects gave written informed consent,which was authorized by the Institutional ReviewBoard.

2.2. Lifting task

Subjects stood comfortably on a force plate infront of a frame with two shelves which werepositioned within reaching limit. The lower shelf was70 cm (desk height) above ground and 40 cm anterior tothe subjects’ ankles. The upper one was 140 cm (shelfheight) above ground and 70 cm anterior to their ankles,respectively.

Subjects lifted and lowered a box 10 times during1min. The lifting cycles were of equal periods, paced bya metronome. The pacing indicated to the subject whento lift the load from the shelf or table, respectively. Thebox was a cube of 25 cm in dimension with two handlesplaced symmetrically 15 cm above the bottom. The boxweighted 4.6 kg, which corresponds to a lifting index of1 according to 1994 NIOSH-Recommended WeightLimits (Waters et al., 1994; Chaffin et al., 1999) and issuggested to ensure a low risk lifting (Waters et al.,1993).

The lifting tasks were performed with two differentstance conditions. In one task subjects stood comfor-tably with their feet in parallel stance. In the other taskthey kept their feet in a step stance position with the leftfoot being placed 20 cm posterior to the right one. Inboth tasks the feet were apart in pelvis width and thesagittal foot axis was externally rotated between 0 and101 (Fig. 1a and b). Before the experiment subjectspracticed the rhythmic of lifting.

2.3. Measurements

We recorded motion by a position and orientationmeasurement system, based on switched magnetic fields(Motion Star with extended range transmitter, Ascen-sion Technology Corporation, Burlington, VermontUSA). Accuracy testing with a grid of predefinedpositions revealed short range accuracy of 4mm(motion o10 cm), in agreement to technical reports(Handbook of Ascension Tech.). Wide range accuracytesting exhibited larger distortions of 2 cm in our setup(motion >10 cm). Position data was sampled at afrequency of 86Hz and low pass filtered (Butterworth,2nd order, fcutoff ¼ 12Hz).

A total of 8 receivers were attached to the trunk,pelvis, legs and arms using double sided tape interfacesand hard foam pads. The trunk marker was placed overthe processus spinosus of the 7th cervical vertebrae. Thepelvis marker was placed at the base of the sacrum. Tworeceivers were placed at the upper arms, midwaybetween acromio-clavicular joint and lateral humeralepicondyle. Another two were placed at the lower armsin the proximal third between the processus olecrani andthe processus styloideus of the ulna. And two receiverswere attached to the thighs, midway between the greatertrochanter and the lateral femoral condyle. All cables ofthe receivers were bundled together on the backs of thesubjects. Recordings during quiet upright stance servedas reference for the calculation of the relative displace-ment of positional data.

A force plate (AMTI, Newton, MA, USA) recordedthree dimensional ground reaction forces (GRF) andmoments. 3D-GRF was sampled at a rate of 100Hz andlow pass filtered (Butterworth, 4th order,fcutoff ¼ 15Hz). COP excursion was calculated in theanterior/posterior and lateral direction from force andmoment data.

A contact switch at the bottom of the box was used toidentify four different phases during the lifting cycle: (1)transition-up, (2) rest-on-shelf, (3) transition-down, (4)rest-on-desk. Each phase was paced for a duration of1.5 s. The signals recorded during the intervals ofcontact between box and desk or shelf were sampledseparately by all measurement systems and used astrigger for synchronization of kinetic and kinematic data.

2.4. Model

The biomechanical model consisted of 8 segments(Fig. 1a and b). The segmental parameters of legs, trunk,upper arms, lower arms and load were calculated bybody mass proportions (Winter, 1990). Each segmentwas calculated as collapsed mass at the approximatedsegmental COM according to geometry and bodydensity distribution. The total body COM was calcu-lated as the weighted sum of the segmental COMs.

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594586

The model was used in two different ways. It waseither driven by the measured positional data of thereceivers to calculated COM displacement for trials ofsubjects or it was used to calculate COP and COM forsimulated lifting in both stance conditions. The simula-tion task was performed with isolated arm motion,without postural responses of trunk and pelvis segments.In the simulation the motion of the box was simplified toa combined vertical and horizontal displacement follow-ing the sinusoidal pattern (Figs. 1 and 2a):

d ¼ t�sinðtÞ ðFigs: 1 and 2aÞ:

The duration of vertical and horizontal displacementwas set to 1 s each with 0.5 s time lag. The simulatedCOM displacement was calculated with influence ofhorizontal transfer of load and arms for each subject(Fig. 2d, stat.). In addition load release profiles duringrest-on-desk and rest-on-shelf were approximated to asinusoidal pattern as revealed by vertical components ofGRF (Fig. 5a and b; Fig. 2d, rel.). This simulated COMdisplacements were used to approximate vertical andhorizontal GRF in a direct forward solution (Winter,1990). The simulated COP displacement includeddynamic components of vertical and horizontal accel-eration of load and arm segments (Fig. 2c and d, d:hor.–d:ver.). As there was no motion of trunk, pelvis or legs,the simulated COM and COP displacements are

identical in both stance conditions under assumptionthat simulated COP displacement does not traveloutside a range of 710 cm in all directions. Therotational dynamics, such as angular accelerations, arenot included in the model.

2.5. Data processing

We analyzed the data of the 4–8th lifting cycles.Within these cycles the pace has stabilized and motionwas rhythmic. Kinetic and kinematic raw data were 100-point time normalized for each transition and rest phase,separately. Then the 400-point data was ensembleaveraged over all subjects. The variability was assessedby calculation of the coefficient of variation (cv) for timeseries data (Winter, 1984).

Overall COP displacement in the anterior and lateraldirection was calculated for the two transition phasesand stance conditions, including simulation. Further-more, within the initial 300ms of the transition phasesthe root mean square (RMS) of horizontal GRFcomponents in the posterior and lateral direction wereassessed. The correlation between trunk and pelvisposition was used to quantify the activation betweenthe two body segments activities. Comparison of stepversus parallel stance condition were performed by twosided t-tests (significant level po0:05).

0

10.8-0.4 -0.2 0 0.2 0.4 0.6

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

vert

ical

dis

plac

emen

t (m

)

horizontal displacement (m)

shelf

trunk

desk

load

arm

leg

0

10.8-0.4 -0.2 0 0.2 0.4 0.6

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

vert

ical

dis

plac

emen

t (m

)

horizontal displacement (m)

shelf

desk

load

arm

trunk

leg

foot

a) b)

parallel step

Fig. 1. Biomechanical models with 8 segments for simulation of a lifting cycle (a) in parallel stance and (b) in step stance. Load trajectory is

approximated in combined vertical and anterior pattern following a ‘t-sinðtÞ’ pattern (black arrows).

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594 587

3. Results

Mean duration of the lifting cycles were equivalent tothe lifting pace of 6 s (Table 1). However, the fourphases within each cycle differed significantly from eachother. The transition phases where significantly longerthan the rest phases. The sum of each transition phaseand the following rest phase was equivalent to thepacing duration of 3 s.

3.1. Kinematic parameters

Fig. 3 shows the displacement pattern vs. time of boththe pelvis and trunk in the sagittal and frontal planes(right=positive). Inspection of the graphs revealed thatsubjects used different strategies for the performance ofthe lifting tasks that were dependent on the base ofsupport. In simulation, no motion of trunk or pelvis waspresent in both stance conditions (Fig. 3a and d).

Intra- and inter-subject variability of the kinematicparameters were generally low (cvintrao0:14;cvintero0:40), except for lateral kinematics, when sub-jects performed the lifts in parallel stance(cvintra;inter > 1).

The sagittal trunk displacement during the lifting taskwas highly congruent for the two stance conditions(Fig. 3b and c) mainly reflecting the primary aspect ofthe task, i.e. moving the box between the desk and shelf.Range of motion was comparable in both stancesituations during transition phases (Table 2). In the

frontal plane, when subjects were in parallel stance,minimal lateral movements of the trunk occurred duringlifting and lowering the load (Fig. 3f). In step stance,however, a distinct four-phasic pattern of the trunkmotion was observed (Fig. 3e).

Both resting phases in both stance conditions showeda tri-phasic trunk displacement pattern in the sagittalplane (Fig. 3b and c). In the frontal plane almost nolateral displacement of the trunk occurred in parallelstance (Fig. 3f). In step stance, however, a distinct bi-phasic pattern occurred (Fig. 3e).

Range of oscillatory pelvic motion in anterior/poster-ior direction was significantly larger in parallel stance ascompared to step stance in both transition phases(Fig. 3b and c; Table 2). In the frontal plane lateralmotion of the pelvis was, absent in the parallel stancecondition. In step stance, however, a distinct pattern ofthe pelvis motion, phase-locked to trunk motionoccurred (Fig. 3e).

A bi-phasic displacement pattern in the sagittal planewas observed during the resting phases. In the frontalplane no lateral displacement of the pelvis occurred inparallel stance. In step stance a bi-phasic pattern wasobserved (Fig. 3e and f).

3.2. Kinetic parameters

COP and COM patterns of the lifting tasks aregiven in Fig. 4. A simulation of the lifting and loweringtask revealed a specific three-phasic pattern of COP

0

0.2

0.4

0.6

d (m

)

a)

-2

0

2

v (m

/s)

b)

up shelf down desk-5

0

5

a (m

/s²)

c)

-202

d.ve

r (c

m)

468

stat

(cm

)

-4-20

rel(

cm)

-202

d.ho

r(c

m)

up shelf down desk

02468

10

sum

(cm

)e)

d)

motion COP reaction

Fig. 2. Patterns of simulated load lifting: components of horizontal (solid) and vertical (dotted) (a) displacement, (b) velocity and (c) acceleration

during a lifting cycle; (d) components of calculated displacement of Center of Pressure related to vertical acceleration (d:ver), horizontal acceleration(d:hor), quasistatic horizontal displacement (stat), and release of load during rest phases (rel); sum of all components (e) exhibits a four-phasic

pattern.

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594588

displacement in both stance conditions (Fig. 4a). Duringthe rest phases the COP displacement revealed abi-phasic pattern. The overall COP displacement waslarger during rest-on-shelf (rangeshelf=4.5 cm) thanduring rest-on-desk phases (rangedesk=2.6 cm). Themodel did not provide COP changes in the frontalplane in both stance conditions (Fig. 4d).

Subjects’ data revealed that the COP pattern isequivalent to the one, predicted by the model, (Fig. 4b

and c). During the transition phases, the overall COPdisplacement was significantly smaller in parallel stanceconditions as compared to step stance condition or aspredicted by the model (Table 3, po0:01). In the frontalplane, COP remained almost unchanged when subjectslifted and lowered the box in the parallel stancecondition (Fig. 4f). In the step stance condition a four-phasic lateral COP displacement pattern occurred, whensubjects lifted and lowered the load (Fig. 4e).

Table 1

Mean duration (S.D.) of the different phases within each lifting cycle given for the parallel and step stance condition separately

Transition-up (s) Rest-on-shelf (s) Transition-down (s) Rest-on-desk (s)

Parallel stance 1.58 (0.15) 1.42 (0.16)* 1.64 (0.19) 1.36 (0.18)*

Step stance 1.60 (0.18) 1.41 (0.19)* 1.59 (0.14) 1.41 (0.17)*

Transition+rest 3.01 (0.07) 3.00 (0.12)

Cycle duration 6.01 (0.08)

*Statistically significant difference between duration of transition phase and following rest phase (po0:05).

< p

oste

rior

(cm

)

ante

rior

>

up shelf down desk

-2

0

2

4

6

parallel

up shelf down desk

-2

0

2

4

6

step

a) b) c)up shelf down desk

-2

0

2

4

6

simulation

up shelf down desk-1

0

1

2

3

4

< le

ft

(cm

)

r

ight

>

up shelf down desk-1

0

1

2

3

4

d) e) f)up shelf down desk

-1

0

1

2

3

4

Fig. 3. Mean displacement of pelvis (solid) and trunk (dashed) in the anterior/posterior and lateral direction: (a, d) simulation of lifting and lifting in

(b, e) step stance and (c, f) parallel stance condition.

Table 2

Amount of oscillatory motion during transition phases for trunk and pelvis

Trunk Pelvis

Phase Transition-up (cm) Transition-down (cm) Transition-up (cm) Transition-down (cm)

Parallel stance 3.6 (3.0) 2.1 (3.5) 0.8 (0.8) 1.2 (1.1)

Step stance 5.0 (3.7) 3.2 (4.0) 0.1 (0.5) 0.3 (0.9)

Comparison n.s. n.s. p ¼ 0:002 p ¼ 0:02

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594 589

During the resting phases a similar COP displacementin the sagittal plane was found during both stanceconditions as predicted by the model (Table 3). Datashowed a bi-phasic COP displacement pattern in theanterior–posterior direction (Fig. 4b and c). In thefrontal plane, step stance was accompanied by a bi-phasic COP displacement, whereas parallel stanceshowed no motion (Fig. 4e and f).

As predicted by the model, COM showed a specificpattern of displacement that was in phase with the COPdisplacement in both stance conditions (Fig. 4a). Theamplitude of COM displacement was smaller than theCOP displacement in all phases (po0:01; Table 3).

As in COP investigation, subjects data revealed thatthe COM pattern was equivalent to that predicted by themodel (Fig. 4e and f). The over all COM displacement

was significantly smaller in parallel stance conditions ascompared to step stance condition or as predicted by themodel (Table 3). Step stance was accompanied by largerlateral COM displacement when compared to parallelstance or simulation (Fig. 4d–f).

Lifting and lowering of the box in parallel stance wasaccompanied by an early negative peak in the sagittalplane shear force which occurred in the first 10–20%(1347130ms after lift-off) of the lifting and loweringphases of the task (Fig. 5c). This force was closelyassociated with the initial backward displacement of thepelvis in the lift-off phases of the task (Fig. 3c). In stepstance, this initial posterior directed peak in shear forcewas almost absent (transition-up) or markedly decreased(transition-down) as compared to the parallel stancecondition (Fig. 5d, Table 4). In the remainder of both

up shelf down desk-4

-2

0

2

4

6parallel

up shelf down desk-4

-2

0

2

4

6step

a) b) c)

simulation

up shelf down desk0

2

4

6

8

10

< p

oste

rior

(cm

)

ante

rior

>

-2

-1

0

1

2

3

up shelf down desk-2

-1

0

1

2

3

up shelf down deskd) e) f)

< le

ft

(cm

)

r

ight

>

-2

-1

0

1

2

3

up shelf down desk

Fig. 4. Mean displacement of COP (solid) and COM (dashed) in the anterior/posterior and lateral direction: (a, d) simulation of lifting and lifting in

(b, e) step stance and (c, f) parallel stance condition.

Table 3

Mean (S.D.) displacement of COP/COM (cm) in anterior/posterior direction. (Note that both parameters are reduced in parallel stance during

transition phases as compared to simulation but not in step stance condition)

Stance Transition-up (cm) Rest-on-shelf (cm) Transition-down (cm) Rest-on-desk (cm)

COP Parallel 6.6 (1.5)* 3.9 (1.9) 6.0 (2.3)* 2.2 (1.4)

Step 9.8 (4.2) 5.3 (2.1) 9.8 (3.6) 2.5 (1.1)

COM Parallel 3.8 (1.4)* 2.9 (0.5) 3.8 (1.5)* 1.3 (0.6)

Step 5.3 (2.0) 3.2 (0.4) 5.1 (2.2) 1.7 (0.9)

*Denotes statistically significant difference to simulation (po0:05).

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594590

transition phases, changes in GRF were significantlylarger in the step stance condition than in the parallelstance condition (Po0:01) indicative of an increasedshear support (Fig. 5c and d). Lateral shear forces wereminimal in parallel stance condition (Fig. 5e). However,lifting in step stance was accompanied by significantlyincreased lateral shear forces as compared to parallelstance (Table 4). It showed a pronounced four-phasicpattern t; closely associated with the acceleration oftrunk and pelvis displacement, respectively (Fig. 5f).

During the resting phases, sagittal shear forcesshowed smaller values than during transition phases(Fig. 5c and d). There were no differences in magnitude

of GRF between the two stance conditions. In thefrontal plane, there were no lateral GRF in the parallelstance condition. Resting in step stance was accompa-nied by a bi-phasic lateral displacement pattern.

3.3. Relationship between the kinematic and kinetic

parameters

When subjects lifted or lowered the load in parallelstance, the spatial coordination of their trunk, pelvisand the resultant COP/COM displacement were out ofphase. The trunk moved in the same direction as theshift of the COP/COM. The pelvis movement, however,

GR

Fve

rtic

al-

BW

[N]

up shelf down desk

parallel

0

50

100

150

a)

0

50

100

150

step

up shelf down deskb)

-15

-10

-5

0

5

10

15

GR

Fan

terio

r[N

]

up shelf down deskc)

-15

-10

-5

0

5

10

15

up shelf down deskd)

up shelf down desk

-10

-5

0

5

10

15

GR

F la

tera

l [N

]

e)

-10

-5

0

5

10

15

up shelf down deskf)

Fig. 5. Mean vertical, sagittal and lateral Ground Reaction Force (7S.D. dotted) of lifting in (a, c, e) parallel stance and (b, d, f) step stance.

Table 4

Mean (S.D.) of Root Mean Square in horizontal GRF (N) during the first 300ms of each transition phase. (Note the increase of anterior/posterior

shear force in parallel stance in transition up and the general increase of lateral shear force in step stance condition)

Anterior/posterior Lateral

Phase Transition-up (N) Transition-down (N) Transition-up (N) Transition-down (N)

Parallel stance 10.3 (4.9) 10.4 (4.4) 2.8 (0.7) 2.9 (0.9)

Step stance 6.7 (2.6) 8.4 (2.9) 5.3 (2.0) 7.3 (2.4)

Difference p ¼ 0:007 n.s. p ¼ 0:001 p ¼ 0:000

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594 591

was in the opposite direction to the trunk and COP/COM (Table 5, negative correlation). No lateraldisplacement of the trunk and pelvis, or of the COP/COM were observed during lifting in the parallel stancecondition.

When subjects performed the tasks in step stance, thetrunk and pelvis moved in phase, especially in thefrontal plane (Table 5, positive correlation). In the initialphase of the lift, the pelvis sagittal motion brieflyopposed that of the COP shift and the trunk displace-ment in the sagittal plane. In the further course of theload transfer, the direction of the pelvis and trunkdisplacement was in phase with the COP/COM shift.Thus, the coordination of hip, trunk displacement andCOP/COM observed for the step stance conditionappeared to fit the inverted pendulum model better,than did the parallel stance condition.

4. Discussion

Evidence based on kinematic and kinetic parametershas been presented which indicates that subjects lifting aload with different base of support use different posturalcontrol strategies. In parallel stance postural responseconsisted of axial movements in the sagittal plane. Suchstrategy was accompanied by increased posterior shearforces after lift-off. In step stance postural responseincluded lateral transfer of the body. Lateral shearforces were increased as compared to parallel stance.

4.1. Model

A stiff ‘Inverted Pendulum Model’ for predicting thepostural restoration following postural perturbationshas been proposed (Johansson et al., 1988). It isapplicable in quasistatic tasks, such as quiet erectstance. The COP displacement determines the COMacceleration (Guersen et al., 1976). Accordingly, thepredicted COP changes cover voluntarily controlledaspects that are related to ankle torque in the sagittal

plane (Winter et al., 1998). It is valid only if externalperturbation is as small as the COM remains within thebase of support (Rietdyk et al., 1999). Other literaturedefines a preferred zone with change of support absent(Popovic et al., 2000). If the COM sways beyond thiszone the ‘Inverted Pendulum Model’ would predict afall, as there is no force to restore balance and steppingis not part of the model (Otten, 1999; Maki andMcIlroy, 1997; McIlroy and Maki, 1993). Hence,stiffness cannot be assumed and multi segmentalmotion, allowing hip movements and accompanyingpostural reactions would be required to explain com-pensation responses (Horak and Nashner, 1986). Thechoice of the ‘Inverted Pendulum Model’ in the presentexperiment was justified as the amount of perturbationscaused by the load and lifting movements was notsufficient to shift the COM outside the preferred zone inthe base of support. A limitation to obtained results maybe that measured motion of critical data (pelvis) is in theorder of long range measurement accuracy. As relativedisplacement data is used exclusively, the better short-range accuracy is applicable. The simplified model (noangular acceleration) may lead to underestimated rangeof simulated COP displacement. The low dynamics inthe simulated task limits the influence of simplification.

4.2. Dependence of postural control on the base of

support

The simulation of the lifting task using the invertedpendulum model did not take into account posturalcontrol efforts. Thus, the COM displacement was aconsequence of the lifting task and the COP displace-ment directly reflects the destabilizing effect of the task.Furthermore, in consensus with previous studies (Nash-ner and McCollum, 1985), the GRF was assumed topoint into the same direction as did the COM, in orderto achieve static balance during the simulation of thelift. Postural adjustments during task execution weredescribed as mechanisms to keep or to restore thesituation where the GRF pointed to the COM (Horak

Table 5

Mean correlation coefficient between trunk and pelvis displacement during the different phases of lifting cycle. A positive correlation coefficient is

indicative of trunk and pelvis motion into the same direction. A negative correlation coefficient is indicative of trunk and pelvis motion into opposing

directions. Values close to ‘0’ represent no correlation and close to ‘1’ or ‘�1’ excellent correlation and anti-correlation, respectively

Direction Stance Lifting phases

Transition-up Rest-on-shelf Transition-down Rest-on-desk

Anterior Parallel �0.76* +0.79* �0.94* +0.78*

Step �0.04 +0.28 �0.46 +0.84*

Lateral Parallel n.a. n.a. n.a. n.a.

Step +0.97* +0.94* +0.97* +0.93*

*Denotes statistically significant correlation between trunk and pelvis (po0:05).n.a. indicates not applicable correlation calculation, due to too low range of motion.

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594592

and Nashner, 1986; Pedotti et al., 1989). Voluntarymovements were possible only by an active torquegeneration around the ankle, that would correspond tothe voluntary activation of the ventral and plantarmuscles around the ankle, an ‘‘ankle strategy’’. Theankle torque produced had to be of sufficient magnitudeto accelerate the COM during lifting or lowering the boxin order to balance compensation when hip and trunkmove in synchrony. The acceleration of the COM isdirectly related to the difference between COP andCOM in quiet standing (Winter et al., 1998). This is notby active or passive control, but is due to puremechanical coupling during stiff posture (Morasso andSchieppati, 1999).

Lifting in parallel stance was associated with anacceleration of the COM, which could not be compen-sated by purely activating the ventral or plantar anklemuscle groups alone. Thus, subjects used an axialpostural compensation strategy, where they pushedtheir pelvis forward and simultaneously moved theirtrunks backward in order to maintain equilibrium.

This actual behavior with the simultaneous activationof both an ankle and opposing hip strategy did not fitthe inverted pendulum model. Instead, subjects initiallyshifted their pelvis backward in order to generatebackward momentum that could be used to initiate thelift-off of the load. This strategy was associated with anearly posterior peak in the GRF (Fig. 5c). These findingssuggest that a lift performed with a short base ofsupport requires a more complex whole body co-activation pattern in order to maintain equilibrium thanthe one observed during step stance. Lifting in parallelstance provided a solid lateral stability as indicated bythe absence of lateral displacement of the body (Fig. 3f).

When subjects lifted with an increased base of supportin step stance, hip and trunk motion were out of phasewith the COP/COM shifts only in the early phase of thelift. During lift-off, subjects appeared to behave morelike a ‘‘crane’’ by utilizing the increased leverageprovided in the sagittal plane by the step stance whileshowing a postural compensation pattern similar to thatwhen lifting in parallel stance. The inverted pendulummodel appeared to fit the behavior during step stancebetter than during parallel stance, at least in theanterior–posterior direction. These findings suggest thatin the anterior–posterior direction the postural controlstrategies required to maintain balance during lift-offwere less complex than in the parallel stance condition.In the further course of the lift performed in step stance,subjects appeared to phase-lock the pelvis and trunk andsimultaneously shift them in the horizontal plane, acrossthe base of support to accomplish the lifting task. In thisphase of the lift, subjects took advantage of theincreased mechanical support, which simplified posturalcontrol since trunk and pelvis were coordinated as oneunit. In this phase of the lift postural compensation may

be sufficiently achieved by muscles around the ankle.However, lateral movement increased in the step stancecondition due to an active lateral shift of the COMbetween the two feet (Fig. 3e). This aspect of thebehavior is comparable to the load transfer in thedouble support phase of gait (Chao et al., 1983;McMahon, 1986). The lateral control of the COM wasassociated with increased lateral shear forces that maymake subjects more susceptible to external lateralperturbations, especially when the load is centered overone leg.

4.3. Implications

Both lifting techniques exhibit positive and negativeaspects. Lifting in step stance reduces complexity ofpostural control in the anterior/posterior direction. But,the task is asymmetric and decreases lateral stabilitywhen compared with lifting in parallel stance. Wecannot recommend either one as being better in termsof postural control.

Acknowledgements

This work was supported by a grant from theVeterans Affairs Research & Development (E 2184-R).Josef Kollmitzer and Gerold Ebenbichler were sup-ported by the ‘‘Erwin Schroedinger Fellowship’’ of theAustrian Science Foundation (Projects: J 1624-MED, J1639-MED).

We would like to thank Professor Carlo J. De Lucafor hosting at the NeuroMuscular Research Center,Boston University, during the research year andProfessor Veronika Fialka-Moser to grant the leave ofabsence from the Department of Physical Medicine andRehabilitation, University of Vienna.

References

Accornero, N., Capozza, M., Rinalduzzi, S., Manfredi, G.W., 1997.

Clinical multisegmental posturography: age-related changes in

stance control. Electroencephalography and Clinical Neurophy-

siology 105, 213–219.

Aruin, A.S., Forrest, W.R., Latash, M.L., 1998. Anticipatory postural

adjustments in conditions of postural instability. Electroencepha-

lography and Clinical Neurophysiology 109, 350–359.

Brown, J.E., Frank, J.S., 1987. Influence of event anticipation on

postural actions accompanying voluntary movement. Experimental

Brain Research 67, 645–650.

Bouisset, S., Zattara, M., 1987. Biomechanical study of the program-

ming of anticipatory postural adjustments associated with volun-

tary movement. Journal of Biomechanics 20, 735–742.

Chaffin, D.B., 1987. Manual materials handling and the biomechanical

basis for prevention of low-back pain in industry-an overview.

American Industrial Hygiene Association Journal 48, 989–996.

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594 593

Chaffin, D.B., 1999. Manual material-handling limits. In: Chaffin,

D.B., Andersson, G.B.J., Martin, B.J. (Eds.), Occupational

Biomechanics, 3rd Edition. Wiley, New York, pp. 315–354.

Chao, E.Y., Laughman, R.K., Schneider, E., Stauffer, R.N.,

1983. Normative data of knee joint motion and ground

reaction forces in adult level walking. Journal of Biomechanics

19, 219–233.

Cordo, P.J., Nashner, L.M., 1982. Properties of postural adjustments

associated with rapid arm movements. Journal of Neurophysiology

47, 287–302.

Crenna, P., Frigo, C., Massion, J., Pedotti, A., 1987. Forward and

backward axial synergies in man. Experimental Brain Research 65,

538–548.

Cresswell, A.G., Oddsson, L., Thorstensson, A., 1994. The influence

of sudden perturbations on trunk muscle activity and

intra-abdominal pressure while standing. Experimental Brain

Research 98, 336–341.

Diener, H.C., Bootz, F., Dichgans, J., Bruzek, W., 1983. Variability of

postural ‘‘reflexes’’ in humans. Experimental Brain Research 52,

423–428.

Dietz, V., Kowalewski, R., Nakazawa, K., Colombo, G., 2000. Effects

of changing stance conditions on anticipatory postural adjustment

and reaction time to voluntary arm movement in humans. Journal

of Physiology 524 (2), 617–627.

Fathallah, F.A., Marras, W.S., Parnianpour, M., 1998. The role of

complex, simultaneous trunk motions in the risk of occupation-

related low back disorders. Spine 23, 1035–1042.

Guersen, J.B., Altena, D., Massen, C.H., Verduim, M., 1976. A model

of standing man for the description of the dynamic behavior.

Agresologie 17, 63–69.

Hodges, P.W., Cresswell, A.G., Daggfeldt, K., Thorstensson, A., 2000.

Three dimensional preparatory trunk motion precedes asymme-

trical upper limb movement. Gait Posture 11, 92–101.

Horak, F., Nashner, L., 1986. Central programming of postural

movements: adaption to altered support-surface configurations.

Journal of Neurophysiology 5, 358–361.

Humphrey, D.R., Reed, D.J., 1983. Separate cortical systems for

control of joint movement and joint stiffness: reciprocal activation

and coactivation of antagonist muscles. Advances in Neurology 39,

347–372.

Ioffe, M., Massion, J., Gantchev, N., Dufosse, M., Kulikov, M.A.,

1996. Coordination between posture and movement in a bimanual

load-lifting task: is there a transfer? Experimental Brain Research

109, 450–456.

Johansson, R., Magnusson, M., Akesson, M., 1988. Identification of

human postural dynamics. IEEE Transactions on Biomedical

Engineering 35, 858–869.

Maki, B.E., McIlroy, W.E., 1997. The role of limb movements in

maintaining upright stance: the ‘‘change-in-support’’ strategy.

Physical Therapy 77, 488–507.

Massion, J., Gahery, Y., 1979. Diagonal stance in quadrupeds: a

postural support for movement. Progress in Brain Research 50,

219–226.

McGill, S.M., Norman, R.W., Cholewicki, J., 1996. A simple

polynomial that predicts low-back compression during complex

3-D tasks. Ergonomics 39, 1107–1118.

McIlroy, W.E., Maki, B.E., 1993. Task constraints on foot movement

and the incidence of compensatory stepping following perturbation

of upright stance. Brain Research 616, 30–38.

McMahon, T.A., 1986. Mechanics of locomotion. In: McMahon,

T.A. (Ed.), Muscles Reflexes and Locomotion. Princeton Uni-

versity Press, New Jersey, pp. 189–233.

Morasso, P.G., Schieppati, M., 1999. Can muscle stiffness alone

stabilize upright standing? Journal of Neurophysiology 83,

1622–1626.

Nashner, L.M., McCollum, G., 1985. The organization of human

postural movements: a formal basis and experimental synthesis.

Behavioral Brain Sciences 8, 135–172.

Nielsen, J., Kagamihara, Y., 1992. The regulation of disynaptic

reciprocal Ia inhibition during co-contraction of antagonistic

muscles. Journal of Physiology 456, 373–391.

Oddsson, L., 1989. Motor patterns of a fast voluntary postural task in

man: trunk extension in standing. Acta Physiologica Scandinavica

136, 47–58.

Oddsson, L., 1990. Control of voluntary trunk movements in man.

Mechanisms for postural equilibrium during standing. Acta

Physiologica Scandinavica, Supplementum 595, 1–60.

Oddsson, L., Persson, T., Cresswell, A.G., Thorstensson, A., 1999.

Interaction between voluntary and postural motor commands

during perturbed lifting. Spine 24, 545–552.

Otten, E., 1999. Balancing on a narrow ridge: biomechanics and

control. Philosophical Transactions of the Royal Society of

London 354, 869–875.

Pedotti, A., Crenna, P., Deat, A., Frigo, C., Massion, J., 1989.

Postural synergies in axial movements: short and long-term

adaptation. Experimental Brain Research 74, 3–10.

Popovic, M.R., Pappas, I.P.I., Nakazawa, K., Kller, T., Morari, M.,

Dietz, V., 2000. Stability criterion for controlling standing in able-

bodied subjects. Journal of Biomechanics 33, 1359–1368.

Rietdyk, S., Patla, A.E., Winter, D.A., Ishac, M.G., Little, C.E., 1999.

Balance recovery from medio-lateral perturbations of the upper

body during standing. Journal of Biomechanics 32, 1149–1158.

Shiratori, T., Latash, M., 2000. The roles of proximal and distal

muscles in anticipatory postural adjustments under asymmetrical

perturbations and during standing on rollerskates. Clinical

Neurophysiology 111, 613–623.

Timmann, D., Belting, C., Schwarz, M., Diener, H.C., 1994. Influence

of visual and somatosensory input on leg EMG responses in

dynamic posturography in normals. Electroencephalography and

Clinical Neurophysiology 93, 7–14.

Toussaint, H.M., Commissaris, D.A.C.M., Hoozemans, M.J.M.,

Ober, M.J., Beek, P.J., 1997. Anticipatory postural adjustments

before load pickup in a bi-manual whole body lifting task.

Medicine and Science in Sports and Exercises 29, 1208–1251.

Toussaint, H.M., Michies, Y.M., Faber, M.N., Commissaris,

D.A.C.M., van Dieen, J.H., 1998. Scaling anticipatory postural

adjustments dependent on confidence of load estimation in a bi-

manual whole-body lifting task. Experimental Brain Research 120,

85–94.

Waters, T.R., Putz-Anderson, V., Garg, A., Fine, L.J., 1993. Revised

NIOSH equation for the design and evaluation of manual lifting

tasks. Ergonomics 36, 749–776.

Waters, T.R., Putz-Anderson, V., Garg, A., 1994. Applications

manual for the revised NIOSH lifting equation, DHHS (NIOSH)

Publication 94–110.

Winter, D.A., 1984. Kinematic and kinetic patterns in human gait:

variability and compensating effects. Human Movement Science 3,

51–76.

Winter, D.A., 1990. Antropometrie. In: Winter, D.A. (Ed.), Biome-

chanics and Motor Control of Human Movement. Wiley, New

York, pp. 51–73.

Winter, D.A., Patla, A.E., Prince, F., Ishac, M., 1998. Stiffness control

of balance in quiet standing. Journal of Neurophysiology 80,

1211–1221.

J. Kollmitzer et al. / Journal of Biomechanics 35 (2002) 585–594594